{ "architectures": [ "StableLmForCausalLM" ], "attention_dropout": 0.0, "bos_token_id": 100257, "eos_token_id": 100257, "hidden_act": "silu", "hidden_dropout": 0.0, "hidden_size": 5120, "initializer_range": 0.01, "intermediate_size": 13824, "layer_norm_eps": 1e-05, "max_position_embeddings": 4096, "model_type": "stablelm", "num_attention_heads": 32, "num_hidden_layers": 40, "num_key_value_heads": 8, "partial_rotary_factor": 0.25, "qk_layernorm": true, "rope_scaling": null, "rope_theta": 10000, "rotary_scaling_factor": 1.0, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.40.0", "use_cache": true, "use_norm_bias": false, "use_parallel_residual": true, "use_qkv_bias": false, "vocab_size": 100352 }